Alternating Least Squares Optimal Variable Weighting Algorithms for Ultrametric and Additive Tree Representations
CLASSIFICATION AS A TOOL OF RESEARCH, W. Gaul and M. Schader (Eds.), Elsevier Science Publishers B.V. (North-Holland), pp. 97-103
7 Pages Posted: 28 May 2016
Date Written: 1986
An alternating least squares algorithm is developed for constructing either an ultrametric or an additive tree representation of profile data. The algorithm simultaneously estimates the best fitting tree and optimal variable weights for converting the profile data into dissimilarities. An application of the algorithm to a set of synthetic data is presented and some extensions of the method are briefly discussed.
Suggested Citation: Suggested Citation